Antiviral Activity of Human Immunodeficiency Virus Type 1 Gag-Specific Cytotoxic T Lymphocyte Targeting Is Not Necessarily Intrinsically Superior to Envelope Targeting

Abstract

Across several cohorts, human immunodeficiency virus type 1 (HIV-1) Gag- and Env-specific CD8⁺ T lymphocyte (CTL) responses have demonstrated inverse and positive correlations, respectively, to viremia. The mechanism has been proposed to be superior antiviral activity of Gag-specific CTLs in general. Addressing this hypothesis, we created two HIV-1 constructs with an epitope translocated from Gag (SLYNTVATL, SL9) to Env, thereby switching the protein source of the epitope. A virus expressing SL9 in Env was similar to the original virus in susceptibility to SL9-specific CTLS. This finding suggests that Env targeting is not intrinsically inferior to Gag targeting for CTL antiviral activity.

Several cohort studies (9-13, 24) have demonstrated a significant inverse correlation between the magnitude and/or breadth of the Gag-specific CD8⁺ T lymphocyte (CTL) response and viremia and often a positive correlation of viremia with the Env-specific CTL response. Because CTL epitope targeting of human immunodeficiency virus type 1 (HIV-1) is an important determinant of antiviral activity (20, 23), it has been hypothesized that Gag-specific CTLs may be generally superior in suppressing HIV-1 replication. Two observations regarding the virus-neutralizing activity of CTLs in vitro suggest that Gag-specific CTLs might be more potent. One study found that simian immunodeficiency virus (SIV) Gag-specific (but not Env-specific) CTLs can kill acutely infected cells very early by recognizing epitopes derived from incoming virions before de novo viral protein translation (15). Another demonstrated that bulk Gag-specific polyclonal CTL primary cell lines had higher levels of antiviral activity against a laboratory strain of HIV-1 than bulk Env-specific cell lines (6). These data have been interpreted to indicate that Gag-targeted CTLs are intrinsically superior to Env-targeted CTLs, perhaps due to a specific protein property such as early epitope presentation (6). Potential caveats to these in vitro experimental findings, however, include the high multiplicity of infection of the target cells in the experimental observation of early killing (15) and the inability to control for lymphocyte effector function and epitope sequence matching (in vivo versus laboratory virus sequences) in the comparisons of in vitro virus suppression by Gag- and Env-specific CTL lines (6).

To explore the role of protein targeting in the antiviral efficiency of CTLs while controlling for lymphocyte function and epitope specificity, we tested the antiviral activity of CTLs targeting the A*02-presented SLYNTVATL epitope (SL9; Gag 77 to 85 in p17) against that of molecular clones of HIV-1 containing this epitope translocated to Env to alter its protein source. This approach allowed us to hold the effector cells constant and to examine whether altering the protein source of the epitope changes the antiviral efficiency of CTLs.

Two molecular clones of HIV-1 NL4-3 were modified to alter the endogenous Gag SL9 epitope to a previously described (4) nonrecognized sequence (Gag-SL9x) and to create the SL9 epitope sequence in either of two locations in Env (Table 1). These Env mutants contained amino acid substitutions to create the SL9 sequence at the Env_813-821 gp41 cytoplasmic domain or the Env_401-409 V4 loop (Env-SL9-V4). These viruses (Gag-SL9x/Env-SL9-gp41 and Gag-SL9x/Env-SL9-V4) were compared to the index NL4-3.1 virus (Gag-SL9/Env-WT), which contains the clade B SL9 consensus sequence (22). Additionally, we used two control viruses, one that contains Gag with the clade B consensus sequence (Gag-SL9) and a second that contains the Gag-SL9x mutation combined with a methionine-to-alanine mutation at position 20 (M20A) of Nef, which selectively neutralizes the downregulation of major histocompatibility complex class I (MHC-I) by Nef (2).

TABLE 1.

Virus constructions used in this study^a

Virus	Gag 76-86 (p17)	Env 399-411 (gp120-V4)	Env 811-823 (gp41)	Nef 20
Gag-SL9/Env-WT (index, NL4-3.1)	RSLYNTVATLY	TWSTEGSNNTEGS	AVNLLNATAIAVA	M
Gag-SL9/Env-WT/Nef-M20A	-----------	-------------	-------------	A
Gag-SL9x/Env-WT/Nef-M20A	-----L--V--	-------------	-------------	A
Gag-SL9x/Env-SL9-gp41	-----L--V--	-------------	-RS-Y-T--TLY-	-
Gag-SL9x/Env-SL9-V4	-----L--V--	-R-LYNTVA-LY-	-------------	-

^aThe SLYNTVATL (SL9) epitope in Gag 77-85 was genetically created in the Env V4 loop of gp120 or the cytoplasmic region of gp41 by substitution mutations within NL4-3, while the native SL9 epitope was altered with T81L/T84V mutations in Gag. Two control viruses also contained the M20A mutation in Nef. Dashes indicate identity with the index virus.

All viruses were compared for their susceptibilities to CTLs, using a previously described assay for suppression of viral replication (1, 4, 20). Briefly, T1 cells were infected at a multiplicity of infection of approximately 10⁻² 50% tissue culture infectious doses (TCID₅₀)/cell and cultured in triplicate in 96-well plates with CTLs at an effector-to-target cell ratio of 1:4. HIV-1 p24 antigen capture enzyme-linked immunosorbent assays (ELISAs) (PerkinElmer) were performed on medium supernatants at days 3 and 6 after infection to assess viral replication. Raw-virus suppression was calculated as the difference in log₁₀ p24 concentrations between control wells without CTLs and those with CTLs at day 6. For normalized comparisons of viral inhibition on day 6, the percent efficiency of log suppression was calculated according to the following formula, as previously described (4): raw log units of virus suppression/log units of p24 in the no-CTL control. In the following inhibition assays, these viruses exhibited similar growth kinetics. Across 10 experiments, the slopes of log₁₀ p24 increase from day 3 to day 6 (without added CTLs) were 0.74 ± 0.11 for the Gag-SL9/Env-WT virus (wild-type NL4-3.1), 0.75 ± 0.17 for the Gag-SL9/Env-WT/Nef-M20A virus (NL4-3.1 with an M20A mutation in Nef), 0.69 ± 0.11 for the Gag-SL9x/Env-SL9-gp41 virus (NL4-3 with the SL9 epitope in Gag knocked out and an M20A mutation in Nef), 0.71 ± 0.14 for the Gag-SL9x/Env-SL9/gp41 virus (NL4-3 with the SL9 epitope in Gag knocked out and SL9 transferred to gp41), and 0.66 ± 0.13 for the Gag-SL9x/Env-SL9-V4 virus (NL4-3 with the SL9 epitope in Gag knocked out and SL9 transferred to the V4 loop of gp120).

An SL9-specific CTL clone (S00001-SL9-1.8) demonstrated modest suppression of the index Gag-SL9/Env-WT virus, which was enhanced in the absence of MHC-I downregulation by Nef (Fig. 1 F; compare A and B), in agreement with prior results (8, 17, 21). The control Gag-SL9x/Env-WT/Nef-M20A (SL9 knockout) virus was not inhibited (Fig. 1C and F), confirming the functional ablation of the native SL9 Gag epitope by T81L and T84V mutations (4). The Gag-SL9x/Env-SL9-gp41 (SL9 epitope in Env gp41) virus exhibited susceptibility similar to that of the index virus (Fig. 1D and F). The Gag-SL9x/Env-SL9-V4 (SL9 in Env V4) virus was also inhibited, although it showed diminished susceptibility (Fig. 1E and F). Results from four repeats of this experiment were normalized as a ratio of the efficiency of log suppression versus that of the index Gag-SL9/Env-WT virus (Fig. 2 A), demonstrating that the similarity between the susceptibility of the index virus and that of the Gag-SL9x/Env-SL9-gp41 (SL9 in Env gp41) virus to the SL9-specific CTLs was consistent. The reason for reduced suppression of the Gag-SL9x/Env-SL9-V4 virus was unclear but may have been due to impaired epitope processing and/or to overall lower expression of Env compared to that of Gag.

FIG. 1.

SL9-specific CTL suppression of HIV-1 with Env containing an SL9 epitope translocated from Gag. T1 cells were acutely infected with the indicated viruses and cultured in the absence or presence of the SL9-specific CTL clone S00001-SL9-1.8. (A to E) Raw p24 values over time are shown; error bars indicate standard deviations for triplicate experiments. (F) The mean efficiency of virus suppression (log₁₀ units of p24 reduction) is plotted for each virus. Each error bar indicates one standard deviation. These data are representative of four experiments (see Fig. 2A).

FIG. 2.

Relative susceptibilities of HIV-1 with Env containing an SL9 epitope translocated from Gag to SL9- and AL9-specific CTLs. The antiviral activities of CTL clones targeting SL9 or AL9 (AIIRILQQL; Vpr_59-67) were tested against the panel of viruses in Fig. 1. The efficiency of virus suppression in each experiment was normalized as a ratio versus the efficiency of suppression of the index virus (Gag-SL9/Env-WT). (A) Means and standard deviations for four experiments with the SL9-specific CTL clone S00001-SL9-1.8. (B) Means and standard deviations for six experiments with the SL9-specific CTL clone S00001-SL9-3.23T. Note that the T cell receptor variable-chain usage of this clone was different from that of S00001-SL9-1.8, although it was derived from the same person (data not shown). (C) Means and standard deviations for three experiments with the SL9-specific CTL clone S00036-SL9-1.9. (D) Means and standard deviations for three experiments with the AL9-specific CTL clone SL00036-AL9-1.1.

To control for CTL susceptibility in general, these viruses were also tested for sensitivity to suppression by an A*02-restricted CTL clone recognizing the AIIRILQQL epitope in Vpr_59-67, S00036-AL9-1.1. Across three separate experiments (Fig. 2B), both Nef-M20A-containing viruses showed increased susceptibility to CTLs regardless of SL9 epitope sequence, consistent with the known effect of Nef on CTL susceptibility (8, 17, 21), and the viruses with SL9 epitope translocations remained susceptible to this Vpr-specific CTL clone at a level similar to that of the index virus containing wild-type Nef.

Overall, these data suggest that the protein origin of the SL9 epitope, whether Gag or Env, does not necessarily have a marked impact on the antiviral efficiency of SL9-specific CTLs. This was observed even though our epitope translocation approach moved the epitope to a less highly expressed protein and could have reduced the processing efficiency of the epitope, which is efficiently processed in situ (7) and therefore immunodominant in chronic infection. Thus, the similar degree of CTL susceptibility of an SL9 Env translocation mutant suggests that Gag targeting is not necessarily intrinsically superior to Env targeting by CTLs. Note that this is consistent with our early studies of Gag- and Env-specific CTL clones, which demonstrated levels of antiviral activity of HLA B*14-restricted clones recognizing the conserved epitope ERYLKDQQL in gp41 that were equivalent or even superior to those of A*02- and B*14-restricted Gag-specific clones (19, 20).

Our results contrast with the demonstration by Sacha et al. that SIV Gag-specific CTLs recognize and kill infected cells much earlier than those targeting Tat or Env, within a few hours after acute infection of target cells (15). It was hypothesized that the plentiful structural Gag protein from incoming virions is processed for epitope presentation on acutely infected cells via the class I pathway, while the Env protein remains on the cell surface after viral membrane fusion to the newly infected cell. In such a case, Gag epitope presentation could occur before de novo HIV-1 protein expression, which requires several steps, including reverse transcription, integration, transcription, and translation. While Sacha et al. demonstrated that Gag- but not Env-specific CTLs could lyse acutely infected cells at a time after infection that was too early for de novo viral protein expression, in the current study we did not see a functional advantage for Gag versus Env targeting for antiviral activity. This may be explained by methodological differences; the early-killing observations were with assays performed under conditions of excess infection of target cells, requiring at least 20 to 100 virions per cell and magnetofection, while our virus suppression assays started with low multiplicities of infection. Consistent with our findings, Vojnov et al. also observed that Gag- and Env-specific CTLs can have similar activities against SIV in similar virus suppression assays (18).

Our results also contrast with those of Chen et al., who found that polyclonal Gag- and Env-specific CTL lines differ markedly in antiviral activity (6). This finding could be due to methodological caveats acknowledged in that study. First, Gag is generally more conserved than Env, and thus it is likely that the epitopes in the laboratory HIV-1 strains utilized in their virus suppression assays better matched the in vivo HIV-1 epitopes for Gag than those for Env. Second, given the key role of antigenic stimulation in the proliferation and differentiation of CTLs, the greater variability of Env could affect CTL effector function in vivo. Anecdotally, our laboratory has found it difficult to derive viable Env-specific CTL clones from the peripheral blood mononuclear cells (PBMCs) of HIV-1-infected persons despite high frequencies detected by enzyme-linked immunospot (ELISPOT) assays.

Given the clinical associations of Gag-specific CTL responses with better immune control of HIV-1 in vivo (9-13, 24), the importance of the inclusion of Gag (and perhaps the exclusion of Env) in CTL-based vaccines has been emphasized. However, our data demonstrate that protein targeting may not be an overriding determinant of CTL efficacy and that Env targeting can be as effective as Gag targeting. This finding underscores the fact that the antigenic unit of CTLs is the epitope, and epitope properties (i.e., the level and timing of presentation by an infected cell, which in turn depend on factors such as processing efficiency [16], HLA binding affinity [5], sequence variability [22], etc.) determine CTL function. Thus, while characteristics of the source protein of an epitope (such as expression level and kinetics) influence its properties, ultimately there are many other factors that determine the final profile of epitope presentation.

The benefit of higher levels of Gag targeting by CTLs in vivo likely reflects general trends in epitope properties conferred by Gag rather than an intrinsic advantage of Gag targeting. Note that the association of Gag targeting with lower viremia is a statistical correlation seen across large numbers of persons; the predictive value of Gag targeting for viremia in an individual is poor (e.g., note Fig. 2a in reference 9). This suggests that Gag-specific CTLs on average are more effective than but not necessarily superior to other CTLs for any given epitope. Because Gag is highly expressed and relatively conserved in sequence, epitope properties such as high expression levels and resistance to escape are candidates for determinants of antiviral efficacy in vivo. Compared to Env, which is expressed at lower levels and is relatively variable in sequence, Gag epitopes should be more highly expressed and more resistant to escape on average, but this is not always the case for any particular Gag and Env epitopes.

Finally, our data do not entirely exclude a role for better antiviral activity of Gag-specific CTL through early recognition of virion-derived Gag epitopes, as observed by Sacha et al. (15). Conceivably, target CD4⁺ T lymphocytes in crowded lymphoid tissues could be infected simultaneously with multiple virions, resulting in enough virion-derived Gag for early epitope presentation and recognition. However, this group also reported recently that Vpr- and Rev-specific CTLs can also kill infected cells before viral protein translation (14), yet CTL responses against Vpr and Rev have not been reported to correlate with better immune control (9-13, 24). If this virion-derived epitope presentation phenomenon does occur in vivo, perhaps factors such as sequence variability offset a benefit of early Vpr and Rev epitope presentation. Another caveat to our results is the artificial nature of epitope translocation from one protein to another; we cannot exclude that we have altered epitope processing and production levels or that we have interfered with some unforeseen property that affects HIV-1 susceptibility to CTLs (as we noted in an earlier study of the translocation of SL9 into Nef, where we inadvertently ablated HLA class I downregulation by Nef [3]). Additionally, the “knockout” of SL9 in Gag could still contribute some degree of recognition, as was seen for clone S00001-SL9-3.23T, which showed some suppression of the control SL9x virus (Fig. 2C).

In summary, we demonstrated that CTLs targeting Gag likely are not intrinsically more effective than those targeting Env in an in vitro model of viral suppression. These data underscore the importance of epitope properties rather than just protein properties for the antiviral activity of CTLs. Thus, for immunopathogenesis studies and vaccine design, it is important to consider CTL targeting in terms of epitopes and their properties rather than being constrained by thinking of whole viral proteins as antigenic units.